{"id":698698,"date":"2025-11-11T16:40:14","date_gmt":"2025-11-11T13:40:14","guid":{"rendered":"https:\/\/buradabiliyorum.com\/en\/a-better-way-of-thinking-about-the-ai-bubble-2\/"},"modified":"2025-11-11T16:40:14","modified_gmt":"2025-11-11T13:40:14","slug":"a-better-way-of-thinking-about-the-ai-bubble-2","status":"publish","type":"post","link":"https:\/\/buradabiliyorum.com\/en\/a-better-way-of-thinking-about-the-ai-bubble-2\/","title":{"rendered":"A\u00a0better\u00a0way of thinking about the AI bubble\u00a0"},"content":{"rendered":"
\n

People often think about tech bubbles in apocalyptic terms, but it doesn\u2019t have to be as serious as all that. In economic terms, a bubble is a bet that turned out to be too big, leaving you with more supply than demand.\u00a0\u00a0<\/p>\n

The upshot: It\u2019s not all or nothing, and even good bets can turn sour if you aren\u2019t careful about how you make them.\u00a0<\/p>\n

What makes the question of the AI bubble so tricky to answer is\u00a0mismatched\u00a0timelines\u00a0between the breakneck pace of AI software development and the slow crawl of constructing and powering a data center.\u00a0<\/p>\n

Because\u00a0these data centers take years to build, a lot will inevitably change between now and when they come online.\u00a0The supply chain that powers AI services is so complex and fluid that\u00a0it\u2019s\u00a0hard to have any clarity on how much supply\u00a0we\u2019ll\u00a0need a few years from now.\u00a0It\u00a0isn\u2019t\u00a0simply a matter of how much people will be using AI in 2028, but how\u00a0they\u2019ll\u00a0be using it, and whether\u00a0we\u2019ll\u00a0have any breakthroughs in energy, semiconductor design,\u00a0or power transmission in the meantime.\u00a0<\/p>\n

When a bet is this big, there are lots of ways\u00a0it can go wrong \u2014 and AI bets are getting\u00a0very big\u00a0indeed.\u00a0\u00a0<\/p>\n

Last week, Reuters reported that an Oracle-linked data center campus in New Mexico<\/a> has drawn as much as\u00a0$18 billion\u00a0in credit\u00a0from a consortium of 20 banks. Oracle has already contracted\u00a0$300 billion\u00a0in cloud services to OpenAI, and the companies have\u00a0joined with\u00a0SoftBank to build\u00a0$500 billion\u00a0in total AI infrastructure as part of the \u201cStargate\u201d project. Meta, not to be outdone, has\u00a0pledged to spend\u00a0$600 billion<\/a>\u00a0on infrastructure over the next three years.\u00a0We\u2019ve\u00a0been tracking all the major commitments\u00a0here\u00a0\u2014 and the sheer volume has made it hard to keep up.\u00a0<\/p>\n

At the same time, there is real uncertainty about how fast demand for AI services will grow.\u00a0\u00a0<\/p>\n

\n
\n

Techcrunch event<\/p>\n

\n

\n\t\t\t\t\t\t\t\t\tSan Francisco<\/span>
\n\t\t\t\t\t\t\t\t\t\t\t\t\t|<\/span>
\n\t\t\t\t\t\t\t\t\t\t\t\t\tOctober 13-15, 2026<\/span>\n\t\t\t\t\t\t\t<\/p>\n<\/p><\/div>\n<\/p><\/div>\n<\/div>\n

A McKinsey survey released last week<\/a>\u00a0looked\u00a0at how top firms are employing AI tools. The results were mixed.\u00a0Almost all\u00a0the businesses contacted are using AI in some way,\u00a0yet\u00a0few are using it\u00a0on\u00a0any real\u00a0scale. AI has\u00a0allowed\u00a0companies to\u00a0cost-cut in specific use cases, but\u00a0it\u2019s\u00a0not making a dent on the overall business. In short, most companies are still in \u201cwait and see\u201d mode. If\u00a0you\u2019re\u00a0counting on those companies to buy space in your data center, you may be waiting a long time.\u00a0<\/p>\n

But even if AI demand is endless, these projects could run into more straightforward infrastructure problems. Last week, Satya Nadella surprised podcast listeners\u00a0by saying he was more concerned with\u00a0running out of data center space\u00a0than running out of chips. (As he put it, \u201cIt\u2019s not a supply issue of chips; it\u2019s the fact that I don\u2019t have warm shells to plug into.\u201d) At the same time, whole data centers are sitting idle because they can\u2019t handle the power demands of the latest generation of chips.\u00a0\u00a0<\/p>\n

While Nvidia and OpenAI have been moving forward as fast as they possibly can, the electrical grid and built environment are still moving at the same pace they always have. That leaves lots of opportunity for expensive bottlenecks, even if everything else goes right.\u00a0<\/p>\n

We get deeper into the idea in this week\u2019s Equity podcast, which you can listen to below.\u00a0<\/p>\n

\n\t